Mike Lambert

Doing a bit of a self-experiment. Goal: use my personal health and travel data to provide personalized diet and exercise recommendations for me, using frontier models but in a way that avoids leaking to them any private information. Strategy: use a local model (Qwen 3.8 Flash Next) to orchestrate, and use powerful remote models as a tool call to benefit from higher-level thinking and knowledge that the local model does not have. Three-layer approach to privacy: Avoid leaking personally identifiable information, or leaking my identity through writing style -> local model writes the queries to the frontier model, not me Avoid leaking who I am through the payment channel -> zkAPI Avoid leaking who I am through networking / IP -> Tor You need all three (and finally we have all three, at least to some extent) A skill file teaches the local model when and how to construct minimally-data-revealing requests to remote models. Use zkAPI-wrapped-with-Tor as a CLI tool. And everything works! I got the recommendations back, info from frontier models helped to improve them. Main deficiencies: * Tor is really not optimized for request-by-request de-linking, which is the only form of network-layer privacy that really makes sense in today's world (long-running identifiers are too fragile). Probably not private enough, and latency 10-100x higher than it could be, at the same time. * The skill file's request construction strategies are definitely far from optimal. * Qwen 3.8 Flash Next is still too slow for comfort. It's comfortably running at 20-30 TPS, but it would only really start to feel fast at 100+ * There is a tradeoff: the more careful you are about what data you give to a remote model, the less it can help you github.com/ethereum/zkapi/pu…

Analyzing the Architecture of Global Trade: An Overview of the CSCGraph Platform Understanding a modern supply chain requires more than just tracking a shipping container. It requires a cross-disciplinary look at corporate ownership, international trade law, industrial production, and historical pricing. CSCGraph* is a platform designed to consolidate these disparate data layers into a single, navigable knowledge graph. The platform currently maps approximately 2.96 million entities and 4.41 million relationships, providing a structured framework for analyzing how goods and capital move across the globe. 1. The Knowledge Graph Framework Unlike standard relational databases that store data in isolated tables, CSCGraph utilizes a graph topology built on Neo4j. This structure is optimized for "path-finding" problems—such as tracing a commodity from a raw material extraction site through various processing stages to its final destination. The graph is categorized into eight primary node types, including: • Trade Entities: 174 countries and over 1,000 global ports. • Economic Classifications: 6,940 commodity codes (HS System) and 839 industrial sectors. • Corporate Data: Nearly 3 million legal entities linked by parent-subsidiary ownership edges. 2. Data Synthesis and Resilience The platform's utility relies on the harmonization of 13 distinct open data sources. By bridging these datasets, the system can link a price spike in the World Bank's "Pink Sheets" to specific bilateral trade flows in UN Comtrade or production shifts in FAOSTAT. A critical component of this architecture is the DBOS (Database Operating System) backend. Because global datasets are massive and prone to timeouts during ingestion, the system uses crash-resilient workflows. Each step of the data transformation is checkpointed to a PostgreSQL database, allowing the system to resume from a point of failure without corrupting the graph state. 3. AI-Driven Investigation To make this volume of data accessible, the platform integrates a specialized Investigation Agent (utilizing Claude 3.5 Sonnet). Rather than requiring the user to write Cypher (graph query language), the agent is equipped with a toolset of 28 functions. This allows the system to: • Compute Market Analytics: Calculate market concentration (HHI) and supply chain dependencies. • Dynamic Mapping: Convert query results into geographic arcs and facility markers using deck.gl. • Automated Visualization: Generate Sankey diagrams and industrial input-output charts based on live graph data. 4. Expanding the Data Horizon While the platform already hosts over 50,000 price records, the roadmap includes a significant expansion of its data intake. Future updates are set to integrate real-time news feeds to correlate geopolitical events with supply chain disruptions, alongside an expanded array of global price indices and environmental satellite data (such as Climate TRACE facility emissions). These additions aim to move the graph from a historical record to a near-real-time monitoring tool. Current Development and Roadmap CSCGraph currently provides a functional baseline for trade, price, and corporate analysis. By providing a unified schema for global economic data, CSCGraph offers a transparent, reproducible method for mapping the complexities of the global market. *This is the current Early Alpha name of the platform and it may not be its final. Disclaimer: All information found here is for informational, entertainment or educational purposes only and should not be construed as personal investment advice. Conduct your own due diligence, or consult a licensed financial advisor before making any investment decisions.